Digital handwriting correction using deep learning

Dong-Hoon Kim, Hyun-Jun Na, Seong-Hun Jeong, Na-Young Kim, Kyeongbo Kong

Abstract: In this digital age, an increasing number of users are adopting tablets to compose writing. Unlike typing on a computer keyboard, it is difficult to correct handwriting automatically on a tablet. This study proposes a unified framework to correct handwriting on a tablet, based on a deep learning framework. First, we developed an optical character recognition and spell checker to correct typo errors in the input handwriting image. Next, we used a convolutional neural network (CNN) to classify the font of the input image. Finally, the corrected text is printed in a classified font. Experimental results provide a similar output to the user's handwriting.